rubybear
BEARS Ruby API Client
Rubybear is an API Client for interaction with the BEARS network using Ruby.
Changes in v0.4.0
- Gem updates
- AppBase Support
- Defaulting to
condenser_api.*
inRubybear::Api
(see below) - Handle/recover from new
AppBase
errors.
- Defaulting to
Rubybear::Stream
now detects if it's stalled and takes action if it has to wait too long for a new block.- Exponential back-off for stalls so that the node doesn't get slammed.
- Short delays (3 times block production) only result in a warning.
- Long delays (6 times block production) may try to switch to an alternate node.
- Fixed internal logging bug that would open too many files.
- This fix also mitigates issues like
SSL Verify
problems (similar to #12)
- This fix also mitigates issues like
- Dropped GOLOS support.
Appbase is now supported.
If you were already using Rubybear::Api
then there is nothing to change. But if you made use of other API classes, like Rubybear::FollowApi
, then the method signatures have changed.
Pre-AppBase:
api = Rubybear::FollowApi.new
api.get_followers('inertia', 0, 'blog', 10)
New Signature:
api = Rubybear::FollowApi.new
api.get_followers(account: 'inertia', start: 0, type: 'blog', limit: 10)
-- or --
Switch to Condenser API:
The other strategy for using this version of Rubybear is to just switch away from classes like Rubybear::FollowApi
over to Rubybear::Api
(also known as Rubybear::CondenserApi
) instead. Then you don't have to update individual method calls.
api = Rubybear::Api.new
api.get_followers('inertia', 0, 'blog', 10)
Note about GOLOS
GOLOS is no longer supported in Rubybear. If you want to continue to use GOLOS, you'll need to branch from v0.3.15 (pre-appbase) and add WebSockets support because GOLOS completely dropped JSON-RPC over HTTP clients support for some reason
Rubybear has never and will never use WebSockets due to its server scalability requirements.
From a client perspective, WebSockets is great. I have nothing against WebSockets. So I might get around to it at some point, but GOLOS won't be part of Rubybear anymore mainly because GOLOS has no plans to implement AppBase.
Changes in v0.3.0
- Gem updates
- Added failover subroutines (see Failover section, below).
- Added method closures support (aka passing a block to yield).
- You can now stream virtual operations (see Streaming section, below).
- Added more documentation.
- Added/expanded more api namespaces:
::BlockApi
,::CondenserApi
,::TagApi
- Addressed an issue with logging on certain Windows configurations.
Fixes in v0.2.3
- Gem updates
- Added low-level support for persistence and retrying API requests.
- Now using exponential back-off for retries.
- Detecting presence of
transaction_ids
(if enabled by the node). - Default for
Hashie
warnings now go to/dev/null
, where they belong. - Added stray methods/operations.
Fixes in v0.2.2
- Gem updates
- Improved support for datatypes and handlers.
- UTF-8 handled more smoothly.
- Simplified operation construction.
- Improved keep-alive defaults.
- Better streaming reliability.
Also see: Documentation
Quick Start
Add the gem to your Gemfile:
gem 'rubybear'
Then:
$ bundle install
If you don't have bundler
, see the next section.
Prerequisites
minimum ruby version: 2.2
Linux
$ sudo apt-get install ruby-full git openssl libssl1.0.0 libssl-dev
$ gem install bundler
macOS
$ gem install bundler
Usage
require 'rubybear'
api = Rubybear::Api.new
api.get_dynamic_global_properties do |properties|
properties.virtual_supply
end
=> "271342874.337 BEARS"
... or ...
require 'rubybear'
api = Rubybear::Api.new
response = api.get_dynamic_global_properties
response.result.virtual_supply
=> "271342874.337 BEARS"
Follower API
api = Rubybear::FollowApi.new
api.get_followers(account: 'inertia', start: 0, type: 'blog', limit: 100) do |followers|
followers.map(&:follower)
end
=> ["a11at",
"abarefootpoet",
"abit",
"alexgr",
"alexoz",
"andressilvera",
"applecrisp",
"arrowj",
"artificial",
"ash",
"ausbitbank",
"beachbum",
"ben99",
"benadapt",
.
.
.
"bearszine"]
Streaming
Here's an example of how to use a streaming instance to listen for votes:
require 'rubybear'
stream = Rubybear::Stream.new
stream.operations(:vote) do |op|
print "#{op.voter} voted for #{op.author}"
puts " (weight: #{op.weight / 100.0}%)"
end
The output would look like this and continue until interrupted.
richman voted for krnel (weight: 100.0%)
rainchen voted for rainchen (weight: 100.0%)
richman voted for exploretraveler (weight: 100.0%)
jlufer voted for michaelstobiersk (weight: 100.0%)
jlufer voted for michaelstobiersk (weight: 100.0%)
patelincho voted for borishaifa (weight: 100.0%)
richman voted for vetvso (weight: 100.0%)
jlufer voted for michaelstobiersk (weight: 100.0%)
richman voted for orcish (weight: 100.0%)
demotruk voted for skeptic (weight: -100.0%)
photorealistic voted for oecp85 (weight: 100.0%)
meesterboom voted for rubenalexander (weight: 100.0%)
thecurator voted for robyneggs (weight: 40.0%)
richman voted for originate (weight: 100.0%)
helikopterben voted for etcmike (weight: 100.0%)
.
.
.
You can also just stream all operations like this:
stream.operations do |op|
puts op.to_json
end
Example of the output:
{
"vote":{
"voter":"abudar",
"author":"rangkangandroid",
"permlink":"the-kalinga-tattoo-maker",
"weight":10000
}
}
{
"vote":{
"voter":"shenburen",
"author":"masteryoda",
"permlink":"daily-payouts-leaderboards-september-16",
"weight":10000
}
}
{
"vote":{
"voter":"stiletto",
"author":"fyrstikken",
"permlink":"everybody-hating-me",
"weight":2500
}
}
{
"comment":{
"parent_author":"mariandavp",
"parent_permlink":"re-onceuponatime-re-mariandavp-the-bridge-original-artwork-by-mariandavp-20160906t182016608z",
"author":"onceuponatime",
"permlink":"re-mariandavp-re-onceuponatime-re-mariandavp-the-bridge-original-artwork-by-mariandavp-20160917t054726763z",
"title":"",
"body":"https://www.bearsimg.com/images/2016/09/17/oldcomputerpics551cb14c.jpg",
"json_metadata":"{\"tags\":[\"art\"],\"image\":[\"https://www.bearsimg.com/images/2016/09/17/oldcomputerpics551cb14c.jpg\"]}"
}
}
{
"vote":{
"voter":"abudar",
"author":"rangkangandroid",
"permlink":"the-journey-north-through-the-eyes-of-kalinga-tradition",
"weight":10000
}
}
{
"limit_order_cancel":{
"owner":"fnait",
"orderid":2755220300
}
}
.
.
.
You can also stream virtual operations:
stream.operations(:producer_reward) do |op|
puts "#{op.producer} got a reward: #{op.coining_shares}"
end
Example of the output:
anyx got a reward: 390.974648 COINS
gtg got a reward: 390.974647 COINS
someguy123 got a reward: 390.974646 COINS
jesta got a reward: 390.974646 COINS
blocktrades got a reward: 390.974645 COINS
timcliff got a reward: 390.974644 COINS
bhuz got a reward: 1961.046504 COINS
.
.
.
Transactions are supported:
stream.transactions do |tx, trx_id|
puts "[#{trx_id}] #{tx.to_json}"
end
Example of the output:
{
"ref_block_num":59860,
"ref_block_prefix":2619183808,
"expiration":"2016-09-17T06:03:21",
"operations":[
[
"custom_json",
{
"required_auths":[
],
"required_posting_auths":[
"acidpanda"
],
"id":"follow",
"json":"[\"follow\",{\"follower\":\"acidpanda\",\"following\":\"gavvet\",\"what\":[\"blog\"]}]"
}
]
],
"extensions":[],
"signatures":[
"2048d7e32cc843adea0e11aa617dc9cdc773d0e9a0a0d0cd58d67a9fcd8fa2d2305d1bb611ac219fbd3b6a77ab60071df94fe193aae33591ee669cc7404d4e4ec4"
]
}
.
.
.
Even whole blocks:
stream.blocks do |bk, num|
puts "[#{num}] #{bk.to_json}"
end
Example of the output:
{
"previous":"004cea0d46a4b91cffe7bb71763ad2ab854c6efd",
"timestamp":"2016-09-17T06:05:51",
"witness":"boatymcboatface",
"transaction_merkle_root":"0000000000000000000000000000000000000000",
"extensions":[],
"witness_signature":"2034b0d7398ed1c0d7511ac76c6dedaf227e609dc2676d13f926ddd1e9df7fa9cb254af122a4a82dc619a1091c87293cbd9e2db1b51404fdc8fb62f8e5f37b4625",
"transactions":[]
}
.
.
.
Transaction Signing
Rubybear supports transaction signing, so you can use it to vote:
tx = Rubybear::Transaction.new(wif: 'Your Wif Here')
vote = {
type: :vote,
voter: 'xeroc',
author: 'xeroc',
permlink: 'piston',
weight: 10000
}
tx.operations << vote
tx.process(true)
You can also post/comment:
tx = Rubybear::Transaction.new(wif: 'Your Wif Here')
comment = {
type: :comment,
parent_permlink: 'test',
author: 'your-account',
permlink: 'something-unique',
title: 'Rubybear Can Post Comments!',
body: 'Yep, this post was created by Rubybear in `ruby`.',
json_metadata: '',
parent_author: ''
}
tx.operations << comment
tx.process(true)
Transfers:
tx = Rubybear::Transaction.new(wif: 'Your Wif Here')
transfer = {
type: :transfer,
from: 'ned',
to: 'inertia',
amount: '100000.000 BSD',
memo: 'Wow, inertia! Rubybear is great!'
}
tx.operations << transfer
tx.process(true)
There's a complete list of operations known to Rubybear in broadcast_operations.json
.
Failover
Rubybear supports failover for situations where a node has, for example, become unresponsive. When creating a new instance of ::Api
, ::Stream
, and ::Transaction
, you may provide a list of alternative nodes, or leave them out to use the default list. For example:
= {
url: 'https://api.bearshares.com',
failover_urls: [
'https://api.bearsharesstage.com',
'https://api.bears.house'
]
}
api = Rubybear::Api.new()
In a nutshell, the way this works is Rubybear will try a node and proceed until it encounters an error, then retry the request. If it encounters a second error within 5 minutes, it will abandon the node and try a random one from failover_urls
.
It'll keep doing this until it runs out of failovers, then it will reset the configuration and go back to the original node.
Rubybear uses an exponential back-off subroutine to avoid slamming nodes when they act up.
There's an additional behavior in ::Stream
. When a node responds with a block out of sequence, it will use the failover logic above. Although this is not a network layer failure, it is a bad result that may indicate a problem on the node, so a new node is picked.
There is another rare scenario involving ::Transaction
broadcasts that's handled by the failover logic: When a node responds with a network error after a signed transaction is accepted, Rubybear will do a look-up to find the accepted signature in order to avoid triggering a dupe_check
error from the blockchain. This subroutine might take up to five minutes to execute in the worst possible situation. To disable this behavior, use the recover_transactions_on_error
and set it to false
, e.g.:
tx = Rubybear::Transaction.new(wif: wif, recover_transactions_on_error: false)
Debugging
To enable debugging, set environment LOG=DEBUG
before launching your app. E.g.:
$ LOG=DEBUG irb -rrubybear
This will enable debugging for the irb
session.
Troubleshooting
Problem: My log is full of Unable to perform request ... retrying ...
messages.
W, [2017-10-10T11:38:30.035318 #6743] WARN -- : database_api.get_dynamic_global_properties :: Unable to perform request: too many connection resets (due to Net::ReadTimeout - Net::ReadTimeout) after 0 requests on 26665356, last used 1507660710.035165 seconds ago :: cause: Net::ReadTimeout, retrying ...
This is caused by network interruptions. If these messages happen once in a while, they can be ignored. Rubybear will retry the request and move on. If there are more frequent warnings, this will trigger the failover logic and pick a new node, if one has been configured (which is true by default). See the Failover section above.
Problem: My log is full of Invalid block sequence
messages.
W, [2017-10-10T13:53:24.327177 #6938] WARN -- : Invalid block sequence at height: 16217674
This is a similar situation to Unable to perform request ... retrying ...
. Rubybear::Stream will retry and failover if needed. It is happening because the node has responded with a block out of order and ::Stream is ignoring this block, then retrying.
Problem: What does the Stream behind
error mean?
W, [2017-10-09T17:15:59.164484 #6231] WARN -- : Stream behind by 6118 blocks (about 305.9 minutes).
Solution:
This is an error produced by ::Stream
when it notices that the current block is falling too far behind the head block. One solution is to just restart the stream and see if it happens again. If you see a message like this occasionally, but otherwise the stream seems to keep up, it probably was able to recover on its own.
There can be several root causes. Resources like memory and CPU might be taxed. The network connection might be too slow for what you're doing. Remember, you're downloading each and every block, not just the operations you want.
If you have excluded system resources as the root cause, then you should take a look at your code. If you're doing anything that takes longer than 3 seconds per block, ::Stream
can fall behind. When this happens, ::Stream
will try to catch up without displaying a warning. But once you fall 400 blocks behind (~20 minutes), you'll start to get the warning messages.
Verify your code is not doing too much between blocks.
Problem: I'm getting an endless loop: #<OpenSSL::SSL::SSLError: SSL_connect SYSCALL returned=5 errno=0 state=error: certificate verify failed>
Solution:
You're probably creating too many threads or you don't have enough resources for what you're doing. One option for you is to avoid persistent HTTP by passing persist: false
.
Doing this will impact performance because each API call will be a separate socket call. All of the constructors accept persist: false
., e.g.:
api = Rubybear::Api.new(persist: false)
... or ...
stream = Rubybear::Stream.new(persist: false)
... or ...
tx = Rubybear::Transaction.new(.merge(persist: false, wif: wif))
Also see troubleshooting discussion about this situation:
https://github.com/bearshares/rubybear/issues/12
Tests
- Clone the client repository into a directory of your choice:
git clone https://github.com/bearshares/rubybear.git
- Navigate into the new folder
cd rubybear
- Basic tests can be invoked as follows:
rake
- To run tests with parallelization and local code coverage:
HELL_ENABLED=true rake
- To run a stream test on the live BEARS blockchain with debug logging enabled:
LOG=DEBUG rake test_live_stream
See my previous Ruby How To posts in: #rubybear #ruby
Get in touch!
If you're using Rubybear, I'd love to hear from you. Drop me a line and tell me what you think! I'm @inertia on BEARS.
License
I don't believe in intellectual "property". If you do, consider Rubybear as licensed under a Creative Commons License.